The situation in the world today has made 5G technology popular for all the wrong reasons. From viruses that travel through radio waves to 5G being a military-grade weapon, the conspiracy theories are endless. Something you can take to the bank, however, is the fact that it’s 10 times faster than 4G and requires a 1ms network latency. Edge computing is pretty much the only way to meet these requirements and probably the reason why we’re seeing several cloud vendors partnering with telecom organizations.
While quite a few different definitions are floating around about edge computing, Gartner’s is probably closest to the money, taking into consideration that this is a pretty hard field to classify. What it boils down to, is mitigating bandwidth limits and decreasing latency by moving resources (compute, storage, and bandwidth) as close to the user (or endpoints) as possible. In today’s world, this refers to cloud computing capabilities and IT service environments that are being moved closer to the edge to facilitate new developments like AR/VR, AI/ML, and 5G, among other things.
AWS on the edge
First on our list of cloud vendors bending over backward to accommodate this new surge in demand for edge computing is none other than the organization that has probably made more money than anyone else during this pandemic. While it’s all documented on the AWS IoT for the Edge site, we’re going to take a quick look at some of the more exciting offerings. The IoT Device Tester is one of those offerings. This is because compatibility is a big issue with the vast number of devices out there. What it is, is a quick and easy way to check for compatibility.
It does this by performing a quick check and informing the user whether a particular device will run FreeRTOS or Greengrass. It also goes a step further to confirm whether the device will also sync up and communicate with other AWS services, or not. Two other major edge offerings from AWS include FreeRTOS, wh that’s an operating system for microcontrollers and embedded MCUs for sensors, and AWS IoT Greengrass, that’s AWS’s IoT edge computing platform that runs on Linux. Greengrass allows Edge devices to act locally on data while still using the cloud for storage and management.
AWS announced earlier this year that its serverless edge offering Lambda@Edge now supports Node 12.x, as well as Python 3.8. AWS has also majorly revamped all the Snowball related offerings to address this growing space. In addition to delivering a 25 percent faster data transfer performance by increasing edge compute capabilities to 40 vCPUs and 80GiB of memory, the new Snowball also supports sbe-c instance types, sports a new GUI and features IAM. Other edgy updates to Snowball include a new automated way to run maintenance and simple tasks on Snowball Edge devices called Snowball Edge Support for AWS Systems Manager.
The Azure Edge
In addition to naming the new replacement for Internet Explorer as Edge, Microsoft is pretty serious about this sector with a number of offerings, including Edge Zone, Stack Edge, and IoT Edge. With an emphasis on helping users accelerate AI applications at the edge, Microsoft earlier this year announced the new T4 Tensor GPU from Nvidia as part of the new configurations for Azure Stack Edge, an AI-enabled, hyperconverged, managed appliance that can run SQL servers and Kubernetes clusters among other things. Azure Stack Edge was previously called Data Box, and like AWS Snowball, it can be used to transfer petabyte-scale data to the cloud quickly and painlessly.
Another interesting offering from Microsoft is Azure IoT Edge that takes features like cloud analytics and custom business logic and packages them in containers to be deployed on edge devices. IoT Edge comprises a cloud-based interface, IoT Edge Modules, the containers your business logic is packaged in, and IoT Edge runtime that runs on and manages each device. In addition to the option to bring your own code, IoT Edge also allows you to deploy AI and machine learning processes without writing any code.
Azure Edge Zone is a collection of Edge offerings from Microsoft that includes Azure Edge Zones, Azure Edge Zones with Carrier, and Azure Private Edge Zones. It’s a scaled-down hardware extension of Azure that can be used outside Azure regions. Use cases include gaming, streaming, AI/ML, rendering, and real-time analytics. The “with carrier” version is meant to be run via mobile operators and is aimed at 5G networks while the private version lets you have an Edge Zone on-premises.
Google rewiring the edge
Not to be outdone or left behind, the search engine giant recently announced a new strategy, as well as a partnership with telecom giant AT&T as part of an initiative for Google’s Global Mobile Edge Cloud. While this is aimed at edge connectivity, especially 5G, this is only a small part of Google’s edge strategy, which even includes designing their own chips. Yes, trust Google to decide CPUs or even GPUs aren’t going to cut it for the edge, and we need new chips.
TPUs or tensor processing units are application-specific integrated circuits (ASIC) designed by Google to run TensorFlow Lite machine learning workloads. Google has two kinds of these chips that complement each other, namely, Cloud TPUs, and the more recent, Edge TPUs. While the cloud TPUs help accelerate ML in the cloud, the Edge TPUs facilitate equally fast ML inference on the edge. Equipped with the Edge TPUs, IoT sensors go beyond just data collection and can make decisions for themselves in real-time.
Google has also announced an edge computing platform called Cloud IoT Edge that lets you make use of the new TPUs, both in the cloud and on the edge. What it is, is a software stack designed to extend Google Cloud capabilities to the edge, especially with regards to data processing, ML, and AI. With the help of this platform, users can not only create and train complex ML models in the cloud but can also run them on IoT and edge devices with the new edge TPUs.
IBM securing its 5G future
Big Blue has been busy creating strategic alliances in preparation for the upcoming transition to 5G. Not only has it partnered with global telecom Verizon on a number of upcoming Edge and 5G projects, but has also teamed up with IT service company Wipro in India for similar reasons. IBM also just recently announced IBM Edge Application Manager, as well as IBM Telco Network Cloud Manager, IBM Edge Ecosystem, and IBM Telco Network Cloud Ecosystem.
While the names are quite long and confusing as with most IBM offerings, the intention to take over the edge is pretty clear. IBM Edge Application Manager is an autonomous solution custom-designed for AI, analytics, and IoT workloads. It’s also the first of its kind to be powered by IBM’s open-source project Open Horizon and allows a single user to manage up to 10,000 edge nodes. The Telco version runs on RedHat OpenShift (which IBM owns) and is aimed at telecom companies looking to gear up for 5G. IBM and RedHat OpenShift are also making their intentions known in the battle for hybrid cloud supremacy with the unveiling of a new 7nm CPU called IBM Power10.
Edge computing creating opportunities and allies
Lastly, IBM Edge Ecosystem and its telco version are both joint efforts between IBM and over 20 different partners, including Cisco, Intel, Nvidia, Samsung, and Dell. They are all working together to implement and maintain open standards on the edge. While the pandemic has indeed sped up the shift to the edge that a lot of us knew was coming, most countries don’t even have 5G yet, so we’re going to probably see a lot of “Game of Thrones” style alliances between telecom and cloud giants, and it’s still anyone’s guess as to who comes out on top.
Featured image: Shutterstock